American Mineralogist, Volume 80, pages 1031-1040, 1995

Zirconosilicate phase relations in the Strange Lake (Lac Brisson) pluton, Quebec-Labrador, Canada

STEFANO SALVI,* ANTHONY E. WILLIAMS-JONES Department of Earth and Planetary Sciences, McGill University, 3450 University Street, Montreal, Quebec H3A 2A7, Canada

ABSTRACT Petrographic observations of subsolvus granites at Strange Lake indicate that the sodium zirconosilicate elpidite crystallized under magmatic conditions, but that the calcium zir- conosilicates armstrongite and gittinsite are secondary. This interpretation is consistent with the extensive solid solution displayed by elpidite and the restricted compositions of armstrongite and gittinsite. Both calcium zirconosilicate minerals show textural evidence of having replaced elpidite, and in the case of gittinsite, with major volume loss. In the near-surface environment, gittinsite plus quartz fill the volume formerly occupied by el- pidite. At greater depth, gittinsite and armstrongite partially replaced elpidite but are not accompanied by quartz, and abundant pore space is observed where gittinsite is the prin- cipal secondary phase. Below 70 m elpidite is generally unaltered. Replacement of elpidite by armstrongite is interpreted to have been a result of the cation-exchange reaction NazZrSi601s'3HzO + Ca2+ = CaZrSi601s.3HzO + 2Na+ elpidite armstrongite in which volume is nearly conserved, and replacement by gittinsite is thought to have resulted from the reaction NazZrSi601s.3HzO + Ca2+ + 5HzO = CaZrSiz07 + 2Na+ + 4H4SiO~ elpidite gittinsite which is accompanied by a 65% volume reduction. An alteration model is proposed in which external Ca-rich, quartz-undersaturated fluids dissolved elpidite and replaced it with gittinsite, where aH4SiO~was buffered mainly by the fluid (high water-rock ratio), and with armstrongite plus gittinsite, or armstrongite alone, where aH4SiO~was buffered to higher values by the rock (low water-rock ratio). The for- mation of gittinsite created extensive pore space that was subsequently filled, in the upper part of the pluton, when the fluid became saturated with quartz as its temperature de- creased during the final stages of alteration.

INTRODUCTION curs as one or more of a large family of alkali and al- In most igneous rocks, Zr is an incompatible trace kaline-earth zirconosilicate minerals. The most impor- element present in amounts ranging from < 10 to several tant of these are catapleiite (NazZrSi309. 2HzO), dalyite hundred parts per million and is accommodated as the (KzZrSi601s), elpidite (NazZrSi601s' 3H20), eudialyte accessory mineral zircon. However, in peralkaline ig- [Na4(Ca,Ce)z{Fe2+ ,Mn,Y)ZrSig022(OH,Cl)zJ, vlasovite neous rocks its concentration is commonly orders of (NazZrSi4011), and wadeite (KzZrSi309), any of which magnitude higher and may be locally sufficient to con- may be the principal zirconosilicate mineral in a partic- stitute a potentially exploitable deposit (e.g., Ilimaussaq: ular setting, e.g., eudialyte at Lovozero (Vlasov et aI., Gerasimovsky, 1969; Lovozero: Kogarko, 1990; Strange 1966), catapleiite at Mont-Saint-Hilaire (Horvath and Lake: Miller, 1986; Kipawa: Allan, 1992; and Thor Lake: Gault, 1990), and vlasovite on Ascension Island (Fleet Trueman et aI., 1988). The Zr mineralogy of peralkaline and Cann, 1967). rocks is also unusual. Although most peralkaline rocks Little is known about the factors controlling the distri- contain some zircon, the bulk of the Zr commonly oc- bution of zirconosilicate minerals, and there is little doc- umentation of their parageneses. The only reversed ex- perimental study that provides data on their stability at ... Presentaddress:Laboratoirede Geochimie-CNRS,Unive!"- geologically meaningful conditions is that of Currie and site Paul Sabatier, 38 rue des Trente-Six Ponts, Toulouse F-31400, Zaleski (1985) on the dehydration reaction of elpidite to France. vlasovite and quartz. Marr and Wood (1992) constructed

0003-004X/94/091 0-1 031 $02.00 1031 1032 SALVI AND WILLIAMS-JONES: ZIRCONOSILICATE PHASE RELATIONS

++++n.r+ + + + +J1L ~ + + + + + + + + ~ + + + + + + + + + . + + + + + + *+ + +

APHEBIAN BASEMENT amphibolite and metagabbro

metadiorite quartzo-feldspathic & graphitic gneiss calc-silicate gneiss

Fig. 1. Location and simplified geological map of the Strange Lake pluton (modified after Salvi and Williams-Jones, 1992). theoretical P-T diagrams for sodium and calcium zir- the calcium zirconosilicate armstrongite, which they conosilicate minerals using the results of Currie and Za- claimed occurs both as phenocrysts and as a replacement leski (1985), molar volume data, and the few constraints of elpidite. available from natural systems. Further advances in our We reexamine the genesis of elpidite, armstrongite, and understanding of Zr mineral petrogenesis will require ba- gittinsite at Strange Lake, in light of textural and mineral sic thermodynamic data for the major phases and careful chemical data collected from a representative diamond- documentation of field occurrences. The latter is the ob- drill hole located near the zone of economic interest. These jective of the present contribution. data, in conjunction with mineral stability relationships In the Strange Lake pluton, Zr occurs mainly in the and published fluid-inclusion data (Salvi and Williams- form of the zirconosilicate minerals elpidite, armstrongite Jones, 1990), support a model involving low-temperature (CaZrSi6015.3HzO), and gittinsite (CaZrSiz07). These metasomatic replacement of elpidite by armstrongite and minerals vary greatly in their relative proportions, and, gittinsite. locally, anyone of them may be the principal or sole Zr phase. The highest concentrations of Zr are in rocks con- GEOLOGICAL SETTING taining only gittinsite. Salvi and Williams-Jones (1990) The Strange Lake pluton (1189 :t 32 Ma, Pillet et aI., observed that gittinsite plus quartz form pseudomorphs 1989), located in the eastern Rae province of the Cana- after an unidentified phase; on the basis of the morphol- dian Shield, is a peralkaline granite that intruded Aphe- ogy of the pseudomorphs and on fluid-inclusion evi- bian metamorphic rocks (1.8 Ga) and an Elsonian quartz dence, they suggested that a Ca-bearing fluid at -200°C monzonite (1.4 Ga) (Fig. 1). The intrusion is a subverti- caused the metasomatic transformation of the sodium cal, cylindrical body approximately 6 km in diameter and zirconosilicate elpidite to gittinsite + quartz. Birkett et represents the latest tectonic event in the area. Large roof aI. (1992) subsequently proposed that gittinsite replaced pendants are commonly observed (Fig. 1), suggesting that SALVI AND WILLIAMS-JONES: ZIRCONOSILICATE PHASE RELATIONS 1033 the present erosional surface is close to the apical part of the pluton. On the basis of its peralkalinity, mode of em- placement, and age, the pluton may represent the Lab- rador extension of the Gardar anorogenic igneous event (Currie, 1985; Pillet et ai., 1989). 18.0 Several concentric intrusive pulses ofpera1kaline gran- ite formed the Strange Lake pluton, represented by hy- persolvus granite, transsolvus granite, and more evolved, 16.0 volatile-saturated subsolvus granites (Nassif and Martin, 1991) (Fig. 1). All phases are F-rich and contain primary fluorite. They are mainly leucocratic, fine to medium 14.0 grained, and quartz saturated. Pegmatites are common and concentrated mainly in the subsolvus granite. An outwardly dipping ring fracture, marked by a heteroge- J.. 12.0 neous breccia in a fluorite-filled matrix, surrounds the N pluton. E The hypersolvus granite consists largely of phenocrysts 0. 10.0 of quartz and perthite, and interstitial arfvedsonite, 0. whereas the subso1vus granite contains two alkali feld- 0 0 8.0 spars (albite and microcline), and arfvedsonite commonly 0 V occurs as phenocrysts. Transsolvus granite is transitional or- LO to the two other types of granite, i.e., it contains perthite ~-II >< phenocrysts in addition to albite and microcline. Arfved- 6.0 II c::: sonite in this unit occurs predominantly as fine-grained c::: anhedra, imparting a melanocratic appearance to the rock. The hypersolvus granite contains up to 5% of zirconosil- 4.0 icate minerals, and the subsolvus granite and pegmatites contain up to 30% of these minerals. Elpidite is the prin- cipal Zr mineral in the hypersolvus granite, whereas el- 2.0 pidite, armstrongite, and gittinsite are the dominant Zr minerals in the subsolvus granite and pegmatites. Less common zirconosilicates at Strange Lake include cata- 0.0 p1eiite, calcium catapleiite, dalyite, hilairite (Na2ZrSi309. 3H20), vlasovite, and zircon (cf. Birkett et ai., 1992). Subsolvus granites commonly display hematitic alter- ation and, where altered, have higher Zr contents than fresh hypersolvus or subsolvus granites (Fig. 2). Hema- Fig. 2. A histogram showing the average whole-rock content tization was associated with the development of second- of Zr in the principal varieties of granite at Strange Lake. Ab- ary high-field-strength element (HFSE) minerals, many breviations: hs = hypersolvus, ts = transsolvus, f-ss = fresh sub- solvus, a-ss = strongly altered sub- of which are Ca-bearing. Some of the most intense he- altered subsolvus, and s-ss = solvus. Standard deviations are 1.9,2.4, 1.3,2.3, and 0.3 x 1000 matization is in the center of the complex, where a sub- ppm, respectively. economic ore zone has been delineated containing some 30 million tonnes grading 3.25% Zr02, 1.3% REE oxides, 0.66% Y203, 0.56% Nb20s, and 0.12% BeO (Iron Ore Williams-Jones, 1992). Significantly, this alteration is most Company of Canada, unpublished data). These granites strongly developed as halos around pegmatites that con- have lower alkali contents, higher FeH /FeH, and higher tain primary high-temperature fluid inclusions rich in Ca contents than fresh or less-altered subsolvus rocks NaCI and without detectable Ca. (Miller, 1986; Salvi and Williams-Jones, 1990; Boily and Williams-Jones, 1994). Salvi and Williams-Jones (1990) GEOLOGY OF DDH SL-182 provided fluid-inclusion evidence of the interaction of The diamond-drill hole that is described in detail be- ore-zone rocks with a low-temperature, CaCl2-dominated low is one of the deeper (88 m) of some 35 drill holes brine. The high 0180 of these rocks suggests that the fluid that were logged by the authors and is located approxi- was externally derived (cf. Boily and Williams-Jones, mately 700 m southeast of the main exploration trench 1994). A second episode of alteration is represented by (Fig. 1). This hole provides a representative cross section the replacement of arfvedsonite by aegirine. This event of the alteration and intersects several varieties of sub- is not well constrained temporally but has been shown to solvus granite, as well as several pegmatites (Fig. 3). All be limited spatially to the subsolvus granite and is thought rock units contain one or more of the zirconosilicate min- to represent an earlier, high-temperature event (Salvi and erals elpidite, armstrongite, and gittinsite. --

1034 SALVI AND WILLIAMS-JONES: ZIRCONOSILICATE PHASE RELATIONS

0 crysts and transsolvus granite inclusions. Subhedral to J: 6 . Q.e o . euhedral aegirine occurs locally. The lower contact be- ~~++ tween this and the porphyritic granite is gradational over 10 + + + + + a few meters, with the proportion of arfvedsonite phe- + + + + nocrysts decreasing upwards and the zirconosilicates hav- + + 20 + + ing two habits (see below). The upper contact is more + + + j + abrupt, although lacking a chilled margin. This and the ++ + + + + absence of other evidence of intrusive relationships sug- 30 + + + "'tJ gest that these two main rock types were roughly coeval. ++ + + + + + o The equigranular granite is intersected in other drill holes 40 + + + and is exposed in the trench, west of DDH SL-182. It + + '" + + + (J)o can therefore be deduced that this unit forms a subhori- ++ +++ ..... 50 zontal sheet a few tens of meters in thickness in the cen- -< tral portion of the pluton (cf. Miller, 1990; Birkett et aI., 1992). 60 The granites are altered to varying degrees, except in the bottom 10m of the hole where they are relatively 70 t fresh. Alteration is manifested as a replacement of arfvedsonite by aegirine :t hematite, replacements in- volving the zirconosilicates (see below), hematite staining 80 (particularly near the top of the hole), red Fe3+-activated cathodoluminescence of feldspars, appearance of second- 90 ary fluorite, and the development of extensive secondary ? porosity. The strongest replacement of arfvedsonite by aegirine is observed in and adjacent to several meterwide 100 pegmatites that cut the equigranular granite.

over-burden elpidite DISTRIBUTION OF ZIRCONOSILICATE MINERALS IN DDH SL-182 pegmatite armstrongite equigranular granite , Euhedral elpidite occurs as doubly terminated ortho- (euhedral elpidite) gittinsite rhombic prisms, with the typical boat shape also ob- porphyritic granite .. served in rocks from Narssarssuk by Goldschmidt (1913). (anhedral elpidite) quartz Euhedral armstrongite has a pseudo-orthorhombic habit indistinguishable from that of elpidite. All observed oc- Fig. 3. A schematic representation of the geology, the zir- currences of armstrongite show polysynthetic twinning. conosilicate mineral habits and distribution, and the porosity Gittinsite occurs almost exclusively as narrow feathery profile of DDH SL-182. Darker shading corresponds to higher crystals forming radiating aggregates. porosity. Elpidite, armstrongite, and gittinsite have distinctly dif- ferent distributions and modes of occurrence (Fig. 3). In The top half and final 13 m of the hole are com posed the porphyritic granite at the bottom of the hole, elpidite of fine- to medium-grained porphyritic granite containing is the principal zirconosilicate mineral, and its abundance phenocrysts of arfvedsonite. This unit contains inclusions ranges from a few percent to 20% of the rock by volume; of transsolvus granite ranging from one to a few tens of armstrongite is absent, and gittinsite is rare. Elpidite oc- centimeters in diameter and constituting 2-5% by vol- curs as anhedra up to 3 mm in diameter interstitial to ume of the rock. feldspar and quartz grains (Fig. 4A) or as large aggregates In the central portion of the hole, the granite is also of small grains «0.1 mm in diameter) also distributed fine to medium grained but lacks arfvedsonite pheno- interstitially to the main rock-forming minerals (Fig. 4B).

- Fig. 4. Photomicrographs ofzirconosilicate mineral textures: ondary zircon spherules and growing in pore space adjacent to (A) (74.9 m depth) anhedral crystal of elpidite (elp) and (B) ag- elpidite; (G) (13.5 m) aggregate offeathery gittinsite crystals plus gregate of small elpidite crystals interstitial to quartz (qtz) and quartz interstitial to quartz and feldspars; (H) (trench sample) feldspar (feld); (C) (54.9 m) euhedral crystal ofelpidite; (D) (53.0 gittinsite-quartz pseudomorph after a euhedral phase. Photos A, m) partial replacement of elpidite by polysynthetically twinned B, C, D, E, and H are in crossed polarized light; F and G in armstrongite (arm); (E) (59.4 m) gittinsite (git) rosettes, accom- plain polarized light. Photos A, B, C, E, and G are 2 mm across; panied by pore space, partly replacing large elpidite crystals; (F) D, F, and H are I mm across. (44.7 m) radiating gittinsite crystals attached to a string of sec- SALVI AND WILLIAMS-JONES: ZIRCONOSILICATE PHASE RELATIONS 1035 1036 SALVI AND WILLIAMS-JONES: ZIRCONOSILICATE PHASE RELATIONS

TABLE 1. Representative and average composition of zirconosilicates

Averages Elpidite Elpidite Armstrongite Gittinsite Near end-member Ca-rich n= 46 n = 14 n= 20 Si~. 62.0 63.5 63.2 63.6 62.9 63.4 41.9 TiO. n.d. n.d. 0.20 n.d. 0.27 0.16 0.17 ZrO. 19.0 18.9 17.9 16.2 17.1 19.2 36.9 FeO' n.d. n.d. n.d. 0.29 0.07 0.07 0.43 MgO 0.60 0.32 0.75 0.54 0.44 0.09 0.23 MnO n.d. n.d. n.d. n.d. 0.07 n.d. 0.74 CaO 0.19 n.d. 1.42 2.43 0.92 10.0 17.7 Na.O 8.83 8.09 7.69 8.04 9.01 n.d. 0.13 K.O n.d. n.d. n.d. n.d. 0.21 0.23 n.d. Nb.Os n.d. n.d. n.d. n.d. 0.46 0.35 0.60 Ce.O, n.d. n.d. n.d. n.d. n.d. n.d. 0.85 Nd.O, 0.46 n.d. n.d. n.d. 0.13 0.10 0.21 Total 91.1 90.8 91.1 91.1 91.6 93.5 99.8

Note: n.d. = not detected. Analyses are reported in weight percent. .Total Fe as FeO.

Elpidite is also the dominant zirconosilicate mineral in Rocks from the exploration trench, similar in texture the equigranular granite (middle of the hole), but it forms and mineral assemblage to the equigranular granite in euhedral crystals (Fig. 4C) up to 3 mm long, and its abun- DDH SL-182, as well as equigranular granite intersected dance varies between 15 and 25% of the rock by volume. at high levels by other drill holes, also contain aggregates Many elpidite crystals display varying degrees of rim and of gittinsite plus quartz and lack porosity. However, in core replacement by armstrongite and gittinsite. The vol- contrast to the upper part of hole DDH SL-182, these ume proportion of elpidite replaced by these minerals aggregates form pseudomorphs after a euhedral phase (Fig. averages 15-20% and increases upwards. Where replace- 4H), which we interpret to be elpidite on the basis of the ment involved armstrongite alone, volume was con- occurrence of elpidite crystals with similar morphology served (Fig. 4D). However, elpidite crystals replaced by in equigranular granite in DDH SL-182. The volume pro- gittinsite or by armstrongite + gittinsite all contain pore portions of gittinsite and quartz in the pseudomorphs, space (Fig. 4E and 4F). Although armstrongite can be whether after euhedral or interstitial elpidite, vary from found in contact with both elpidite and gittinsite, the lat- 40 to 70% and 30 to 60%, respectively. Interestingly, the ter minerals rarely occur in contact with each other. In gittinsite-quartz pseudomorphs in both equigranular and the transitional zone between the two granites, both in- porphyritic granite occur where hematite alteration is most terstitial and euhedral habits of elpidite are observed. intense. Farther up the hole (15-46 m), in porphyritic granite, elpidite has a habit similar to that of elpidite in the deep- Geochemistry est part of the hole. However, in this interval it experi- Electron microprobe analyses were performed with a enced extensive dissolution, leaving behind pore space Cameca CAMEBAX equipped with an EDS detector (up to 10% of the rock) and calcium zirconosilicates, (NORAN high-purity Ge detector, NORV AR window). largely gittinsite (Fig. 4F). In cases where an entire pocket Analyses were conducted in EDS mode to permit use of of elpidite was replaced, the proportions of gittinsite and a low beam current (15 kV, 0.6 nA current, and 2000- pore space are about 40 and 60%, respectively, of the 3500 counts), to minimize the volatility oflight elements precursor elpidite. Armstrongite is a minor phase and such as Na (light-element volatility is the major source replaced either cores or rims of elpidite crystals. Replace- of error in zirconosilicate mineral analysis; see Bonin, ment of elpidite is more extensive toward the top of the 1988; Birkett et aI., 1992). Where possible, the beam was interval. defocused to a diameter of 5 ,urn to minimize further In the top 10-15 m of drill core, gittinsite is the sole light-element volatility. Data reduction was performed zirconosilicate mineral. It forms feathery radiating crys- with the PROZA version of the rf>(p,Z) correction routine tals up to 0.5 mm in length embedded in tiny anhedral using internal standards (Bastin and Heijligers, 1991). quartz grains. Gittinsite and quartz form coherent aggre- Compositions of each zirconosilicate are listed in Table gates in the groundmass, interstitial to other minerals (Fig. 1 and are plotted as atomic proportions of Na, Ca, and 4G). The sizes, shapes, and sharp boundaries of these Zr/(Zr + Si) in Figure 5. The data indicate that elpidite aggregates suggest that they fill interstitial volumes for- contains up to 37 mol% of armstrongite in solid solution. merly occupied by crystals of elpidite. In this interval, By contrast, most of the gittinsite and armstrongite sam- gittinsite makes up from 5 to 20% of the rock by volume. ples have compositions very close to those of the ideal The granite is almost free of porosity and displays a patchy end-member phases. Generally, cores of elpidite euhedra red coloration because of fine-grained hematite in the git- are more calcic than the rims and fine-grained, late-in- tinsite-quartz aggregates. terstitial aggregates such as shown in Figure 4B. SALVI AND WILLIAMS-JONES: ZIRCONOSILICA TE PHASE RELA nONS 1037

Zr / (Zr+Si) to be affected by the electron-beam power density. Spec- tra ofarmstrongite are dominated by a band at about 415 r nm and by a hump between 500 and 650 nm (Fig. 6B). gittinsite The band at 415 nm closely resembles Eu2+ activation in strontianite, apatite, and synthetic fluorite described by Mariano and Ring (1975). We therefore interpret Eu2+ to be the principal activator of luminescence in armstron- gite; other lanthanides may account for the hump at 500- 650 nm. Na 10 20 80 90 Ca Gittinsite has a distinct bright orange luminescence. This color is due to a very strong band centered at 600 Fig. 5. Cation proportions in the Strange Lake zirconosilicate nm (Fig. 6C). A wide band is also present at about 420 minerals calculated on the basis of seven and 15 atoms of 0 for nm but is partially eclipsed by the 600 nm peak. Mn ion anhydrous and hydrous species, respectively. Solid circles indi- (Mn2+) activation in apatite, feldspar, and carbonates has cate end-member compositions. been shown to produce emissions between 520 and 650 nm, depending on the Mn site configuration (Marshall, Cathodoluminescence 1988; Mason and Mariano, 1990). Microprobe analyses indicate that gittinsite contains up to 1.5 wt% MnO, sug- Elpidite, armstrongite, and gittinsite display strong lu- gesting that Mn2+ may be the activator in this mineral. minescence when excited by an electron beam. Similar Activation at 420 nm is similar to the homologous emis- but weaker luminescence is observed under ultraviolet sion in armstrongite and elpidite and is therefore attrib- illumination. This property allowed the distribution of uted to Eu2+. zirconosilicates to be established in many drill holes. Cathodoluminescence observations were made using an ELM-3R Luminoscope (Herzog et aI., 1970). Emission DISCUSSION spectra were collected in the 350-850 nm wavelength The euhedral habit of elpidite in the equigranular gran- range, with a grating-type H-20 Jobin-Yvon spectrometer ite suggests that this mineral crystallized relatively early, equipped with an R-928 photomultiplier detector (Mari- whereas its occurrence as interstitial anhedra in the por- ano and Ring, 1975). phyritic granite indicates relatively late-stage crystalliza- Elpidite typically displays a greenish luminescence, tion. In both rock types, however, elpidite is clearly a which generally decays to a duller shade of blue-green magmatic mineral and was the first zirconosilicate to form. after 2-3 min of continued exposure to the electron beam. By contrast, the textures described above show that arm- To minimize this, spectra for elpidite were collected with strongite and gittinsite occur only as secondary subsoli- a defocused beam, reduced power, and increased scan- dus minerals, largely as replacements of elpidite. ning interval. Emission spectra of elpidite show well-de- The formation of armstrongite is readily explained by fined peaks at 425,485, 550, and 580 nm (Fig. 6A). From the cation-exchange reaction comparison with spectra in Marshall (1988), we interpret Na2ZrSi601s.3H20 + Ca2+ the peaks at 485 and 580 nm to represent activation by elpidite Dy3+; the 425 and 550 nm peaks are tentatively assigned to activation by Eu2+ and Tb3+. Armstrongite luminesces hues of blue and does not seem armstrongite

40 Tb" 80 Eu'. 160 :Mn" DYf : GITTINSITE ARMSTRONGITE Dy"' ;>, ;>, Eu" i ELPIDITE .~ 60 Dy" ? :::: 120 .~ 30 ,:: ,:: := '" " .....<= (\ .9" .9" 40 20 .::" 80 .::" Fe" OJ .~" ] .::= ~20 " 10 " ~" 40

o o o 300 400 500 600 700 800 300 400 500 600 700 800 300 400 500 600 700 800 A Wavelength (nm) B Wavelength (nm) C Wavelength (nm) Fig. 6. Cathodoluminescence emission spectra for zirconosilicate minerals (A) elpidite, (B) armstrongite, and (C) gittinsite in subsolvus granites from Strange Lake. Beam voltage: 10-15 keV; beam current: 0.5-1 mA; beam diameter: 5 mm. Scanning at 0.5 nm increments, stepping time of I s. Photomultiplier high-voltage: 950 V. -.---

1038 SALVI AND WILLIAMS-JONES: ZIRCONOSILICATE PHASE RELATIONS

This reaction has a very small negative.:l V, (-11.01 cm3/ The mineralogical relationships in DDH SL-182 occur mol; calculated from molar volume data in Marr and throughout altered parts of the pluton and suggest that an Wood, 1992), which corresponds to 4.8% volume reduc- episode of Ca metasomatism caused extensive replace- tion and is consistent with the textural evidence of essen- ment of elpidite by calcium zirconosilicates to a depth of tially constant volume replacement of elpidite euhedra -70 m below the present erosional surface, and that silica by armstrongite (Fig. 4D). was added in the form of quartz closer to the surface. On The replacement of elpidite by gittinsite can be ex- the basis of data from primary fluid inclusions in gittin- plained by the reaction site-quartz pseudomorphs, Salvi and Williams-Jones (1990) proposed that pseudomorphism was effected by Na2ZrSi601s" 3H20 + Ca2+ + 5H20 Ca-rich brines at temperatures of -200°C. The replace- elpidite ment of magmatic elpidite by gittinsite or armstrongite or both was probably caused by a similar fluid. Such gittinsite low-temperature formation of gittinsite and armstrongite explains their near-end-member compositions. This con- Reaction 2 has a large negative .:lV, (-149.1 cm3/molor trasts with the extensive solid solution displayed by el- 65% volume reduction), thereby explaining the consid- pidite (cf. Fig. 5), which is interpreted to have crystallized erable porosity associated with replacement of elpidite by under magmatic conditions; magmatic conditions for el- gittinsite in the central part of the hole (Fig. 4E and 4F). pidite crystallization are also suggested by zoned crystals In fact, the pore volume predicted by the.:l V, of Reaction that, like plagioclase in igneous systems, were initially Ca 2 compares closely with the observed porosity created rich. The activation of luminescence by Eu2+ in both where grains of elpidite have been entirely replaced by armstrongite and gittinsite but not elpidite (Fig. 6) is con- gittinsite: 60-70%. The production of silicic acid in Re- sistent with Ca metasomatism, and the fact that, in its action 2 and the absence of quartz accompanying gittin- divalent state, Eu readily substitutes for Ca. site indicates that the altering fluid was quartz undersat- The questions that now need to be addressed are urated. whether the Ca metasomatism was caused by orthomag- In the porphyritic granite in the top 10 m of the hole, matic fluids (Birkett et aI., 1992) or externally derived as well as in equigranular granite at shallow levels in oth- formational waters (Salvi and Williams-Jones, 1990); er holes and exposed in the trench, elpidite is absent, whether or not Zr was mobilized and concentrated by the gittinsite is the sole zirconosilicate, and there is no evi- metasomatism; why armstrongite formed at depth and dence of porosity. Instead, quartz fills the interstices gittinsite closer to the surface; why quartz addition was around gittinsite. The similarity in the sizes and shapes restricted to the near-surface environment; and ultimate- of the gittinsite-quartz aggregates and the boat-shaped ly under what conditions and in response to what factors pseudomorphs to elpidite aggregates and euhedra in sim- alteration of elpidite to armstrongite and gittinsite oc- ilar rocks deeper in DDH SL-182 suggest that gittinsite curred. and quartz formed at the expense of precursor elpidite. The orthomagmatic fluid hypothesis can be ruled out This requires that gittinsite replacement of elpidite was by the evidence that the fluids responsible for the deeper accompanied or followed by precipitation of additional alteration were quartz undersaturated. If the altering fluid quartz. A reaction that might explain coprecipitation of were of magmatic origin it would have been in equilib- gittinsite and quartz is rium with the granite and therefore saturated with respect Na2ZrSi601s"3H20 + Ca2+ to silica (any cooling would have caused deposition of elpidite quartz). Moreover, the average Ca concentration ob- served in altered subsolvus granite (1.5 wt% vs. 0.5 wt% in unaltered granites) cannot be explained by magmatic gittinsite quartz processes alone (cf. Nassif, 1993; Boily and Williams- However, .:lV, for this reaction is - 58.4 cm3/moi. This Jones, 1994). represents a 25% volume reduction, which should be The restriction ofgittinsite mainly to the volume of the readily evident as pore space. The fact that there is no precursor phase may indicate that Zr acted as an immo- unfilled volume in the space occupied by gittinsite-quartz bile element during metasomatism. However, as reported aggregates indicates that, if replacement was according to above, the proportions of gittinsite to quartz in samples Reaction 3, there must have been precipitation of addi- from high levels are generally greater than the propor- tional quartz. An alternative explanation, supported by tions of gittinsite to pore space lower in the hole. Al- preservation of delicate rosettes of gittinsite, is that re- though the difference is small, it could be significant be- placement took place by Reaction 2 and quartz precipi- cause it may indicate that Zr was indeed remobilized by tated later. This implies that the metasomatic fluid was the metasomatic fluid, albeit to a limited degree. This initially undersaturated with respect to quartz, and that would be consistent with the observed increase of whole- at a later stage, silica solubility decreased thereby causing rock Zr in the strongly altered subsolvus granites (Fig. 2), precipitation of quartz in the pores at high levels in the i.e., in rocks containing the gittinsite-quartz pseudo- pluton. morphs. SALVI AND WILLIAMS-JONES: ZIRCONOSILICATE PHASE RELATIONS 1039

feathery gittinsite crystals embedded in quartz, we prefer the latter explanation and propose that quartz was pre- cipitated in response to cooling of the fluids during the waning stages of the hydrothermal system (Fig. 7C).

ACKNOWLEDGMENTS The research described in this paper was supported by NSERC strategic and operating grants awarded to A.E.W.-J. and forms part of a larger project involving S. Aja, M. Boily, R. Marr, R. Martin, G. Nassif, J. A Roelofsen-Ahl, and S. Wood. Additional funding was provided by FCAR (Quebec) and a NATO travel grant. We are indebted to the Iron Ore Company of Canada for allowing access to the Strange Lake property and their drill core library. The manuscript benefited from critical reviews by

ELPfDlTE F. Dudas, J Gittins, and an anonymous reviewer.

+ + B REFERENCES CITED z'" '"u'"

"'ro 1 ro Allan, J.F. (1992) Geology and mineralization of the Kipawa yttrium- bt:J o zirconium prospect, Quebec. Exploration and Mining Geology, 1,283- .- ARMSTRONGITE 295. Bastin, G.F., and Heijligers, HJ.M. (1991) Quantitative electron probe microanalysis of ultra-light elements (oxygen-boron). In K.F.J. Hein- rich and D.E. Newbury, Eds., Electron probe quantitation, p. 145-162. Plenum, New York. Fig. 7. Schematic log aF.../ac.'+ vs. aH,SiO: diagrams showing Birkett, T.C., Miller, R.R., Roberts, AC., and Mariano, AN. (1992) Zir- the stability fields of the zirconosilicate minerals, as well as hy- conium-bearing minerals of the Strange Lake intrusive complex, Que- pothetical compositions of the unaltered elpidite-bearing granite, bec-Labrador. Canadian Mineralogist, 30,191-205. the hydrothermal fluid (A), and the altered gittinsite-bearing Boily, M., and Williams-Jones, A.E. (1994) The role of magmatic and granite (C). Also shown are possible paths of metasomatic trans- hydrothermal processes in the chemical evolution of the Strange Lake formation of the rock during alteration (B). The shaded fields plutonic complex (Quebec-Labrador). Contributions to Mineralogy and represent quartz saturation levels in the fluid during metaso- Petrology, 118, 33-47. matism (A and B) and at lower temperature (C). Bonin, B. (1988) Peralkaline granites in Corsica: Some petrological and geochemical constraints. Rendiconti della Societa Italiana di Mineralo- gia e Petrologia, 73, 1191-1194. Currie, K.L. (1985) An unusual peralkaline granite near Lac Brisson, Que- Further insight into this metasomatic process is gained bec-Labrador. Geological Survey of Canada Report 85-IA, p. 73-90. by examining the stability fields of the zirconosilicate Currie, K.L., and Zaleski, E. (1985) The relative stability of elpidite and vlasovite: A PoT indicator for peralkaline rocks. Canadian Mineralo- minerals as a function ofNa, Ca, and Si activity. Unfor- gist, 23, 577-582. tunately, there are few thermodynamic data for the alkali fleet, S.G., and Cann, JR. (1967) Vlasovite: A second occurrence and a and alkaline-earth zirconosilicates (cf. Marr and Wood, triclinic to monoclinic inversion. Mineralogical Magazine, 36, 233- 1992). Nonetheless, it is possible to represent phase-sta- 241. bility relationships schematically using the principles of Gerasimovsky, V.I. (1969) Geochemistry of the Ilimaussaq Alkaline Mas- sif (South-West Greenland), 174 p. Nauka, Moskow (in Russian) (not chemographic analysis (Zen, 1966). In Figure 7 we pre- seen; extracted from Lithos, 26,167,1990). sent a schematic log a~.+/ac.2+vs. log aH,SiO:diagram for Goldschmidt, V. (1913) Atlas der Krystallformen, vol. 3. Carl Winters the system NazO-CaO-ZrOz-SiOz-HzO and the phases in Universitatsbuchhandlung, Heidelberg, Germany. the system observed in the altered rocks of Strange Lake, Herzog, L.F., Marshall, DJ., and Babione, R.F. (1970) The luminoscope: plus vlasovite. Also shown on the diagram are possible A new instrument for studying the electron-stimulated luminescence of terrestrial, extra-terrestrial and synthetic materials under the micro- locations of the boundary for quartz saturation. scope. In J.N. Weber and E. White, Eds., Space science applications of We propose that, sometime after crystallization of at solid state luminescence phenomena, p. 79-98. MRL publication 70- least the apical region of the intrusion, heated ground- 10 I, State College, Pennsylvania. waters in equilibrium with adjacent calc-silicate gneisses Horvath, L., and Gault, R.A (1990) The mineralogy of Mont Saint-Hi- laire, Quebec. Mineralogical Record, 21, 284-359. and gabbros (and therefore Ca-rich and quartz undersat- Kogarko, L.N. (1990) Ore-forming potential of alkaline magmas. Lithos, urated) entered the pluton (Fig. 7A). In the uppermost 26, 167-175. parts where water-rock ratios were high, the fluids dis- Mariano, AN., and Ring, PJ. (1975) Europium-activated cathodolumi- solved elpidite, replacing it primarily with gittinsite (Fig. nescence in minerals. Geochimica et Cosmochimica Acta, 39, 649- 7B, path I). Deeper in the pluton, fluid-rock ratios were 660. Marr, R.A., and Wood, S.A. (1992) Preliminary petrogenetic grids for lower, permitting the rock to buffer aH,SiO:to higher val- sodium and calcium zirconosilicate minerals in felsic peralkaline rocks: ues, which were sufficient to stabilize gittinsite plus arm- The SiO,-Na,ZrO, and SiO,-CaZrO, pseudobinary system. American strongite and locally armstrongite alone (Fig. 7B, path II). Mineralogist, 77, 810-820. The presence of quartz with gittinsite near the surface Marshall, DJ. (1988) Cathodoluminescence of geological materials, 146 p. Unwin-Hyman, Boston. requires either that P- T conditions of alteration changed Mason, R.A, and Mariano, AN. (1990) Cathodoluminescence activation so as to permit saturation of quartz at lower aH,SiO:or that in manganese-bearing and rare earth-bearing synthetic calcites. Chem- quartz was introduced later. In view of the textures, i.e., ical Geology, 88, 191-206. 1040 SALVI AND WILLIAMS-JONES: ZIRCONOSILICATE PHASE RELATIONS

Miller, R.R. (1986) Geology of the Strange Lake Alkalic Complex and the bee/Labrador: Evidence from fluid inclusions. Geochimica et Cosmo- associated Zr-Y-Be-REE mineralization. Newfoundland Department chimica Acta, 54, 2403-2418. of Mines and Energy, Mineral Development Division Report 86-1, p. -(1992) Reduced orthomagmatic C-O-H-N-NaCi fluids in the 11-19. Strange Lake rare-metal granitic complex, Quebec/Labrador, Canada. - (1990) The Strange Like pegmatite-aplite hosted rare metal de- European Journal of Mineralogy, 4, 1155-1174. posit, Labrador. Newfoundland Department of Mines and Energy, Trueman, D.L., Pedersen, J.e., de St. Jorre, L., and Smith, D.G.W. (1988) Geological Survey Branch, Report 90-1, p. 171-182. The Thor Lake, N.W.T., rare-metal deposits. In R.P. Taylor and D.F. Nassif, GJ. (1993) The Strange Lake peralkaline complex, Quebec-Lab- Strong, Eds., Granite-related mineral deposits: Geology, petrogenesis rador: The hypersolvus-subsolvus granite transition and feldspar min- and tectonic setting. Canadian Institute of Mining and Metallurgy, Spe- eralogy, 104 p. M.Sc. thesis, McGill University, Montreal, Quebec. cial Volume 39,279-284. Nassif, GJ., and Martin, R.F. (1991) The hypersolvus granite-subsolvus Vlasov, K.A., Kuz'menko, M.Z., and Es'kova, E.M. (1966) The Lovozero granite transition at Strange Lake, Quebec-Labrador. Geological As- Alkali Massif, 627 p. Oliver and Boyd, Edinburg. sociation of Canada Abstracts with Program, 16, A89. Zen, E-An (1966) Construction of pressure temperature diagrams for mul- Pillet, D., Bonhomme, M.G., Duthou, J.L., and Shenevoy, M. (1989) ticomponent systems after the method of Schreinemakers: A geomet- Chronologie Rb/Sr et K/ Ar du granite peralkalin du lac Brisson, Lab- rical approach. U.S. Geological Survey Bulletin, 1225, I-56. rador central, Nouveau-Quebec. Canadian Journal of Earth Sciences, 26, 328-332. Salvi, S., and Williams-Jones, A.E. (1990) The role of hydrothermal pro- MANUSCRIPT RECEIVED AUGUST 12, 1994 cesses in the granite-hosted Zr, Y, REE deposit at Strange Lake, Que- MANUSCRIPT ACCEPTED MAY 31, 1995